All geek questions in one place

A simple algorithm for optimizing particle pianos in 1-dimensional mode for a noisy environment

I am experimenting with particle rum optimization and trying to determine the best approach for the following simple scenario:

  • Optimization of a one-dimensional function (i.e. particles moving along one line)
  • An optimized function can be selected anywhere on the line.
  • The “value” selected for each position is very noisy.
  • The main optimization function (excluding noise) is quite simple (for example, a pyramid with one global maximum or two humps with different heights).

What would be the best particle rhombus design to solve this problem, that is, to efficiently find the optimum with a minimum number of samples?

+3
source share
2 answers

I don't think Swarm Particle optimization is right for your case ...

PSO retains the maximum value found so far and its location. Therefore, if this highest value was caused only by noise , it will depend there all the time.

You have to modify the regular PSO to consider the weighted average of the found maxima (and not the final global maximum)

I recommend you take a look at: Fuzzy Particle Blur Optimization But this question is still under investigation, so it may be difficult for you to get to it ...

You can also go to Simulated Annealing , which is best suited to your business ...

+2

, , y .

+1

Source: https://habr.com/ru/post/1760990/

More articles:


All Articles